Lead/Senior Research Scientist – Agentic Reinforcement Learning
Core
Lead research on agentic reinforcement learning (RL) to train frontier-scale LLMs for complex, multi-step tasks in professional domains like legal, tax, and compliance.
Role type
Senior IC research scientist (agentic RL)
Builds
Frontier-scale LLMs with agentic behavior (planning, reasoning, tool use) for professional domains
Domain
AI/ML, Reinforcement Learning, Legal/Tax/Compliance
Deliverable
production ML models
Required skills
RL for LLMs, reward modelling, policy optimisation, PyTorch, research leadership, experimental design, mentorship
Preferred skills
NeMo-RL, verl, OpenRLHF, TRL, Mixture-of-Experts, long-context models, knowledge-intensive tasks
Technologies
PyTorch, NeMo-RL, verl, OpenRLHF, TRL
Responsibilities
Lead research agenda in agentic RL and reward design; Invent reliable training signals for real-world tasks; Design ablations and diagnostics; Publish at top-tier venues; Mentor scientists and engineers; Communicate findings via seminars and publications
Seniority
Senior, hands-on IC with mentorship